Impressed By Claude 3.5's Coding Ability

I've been testing out various AI coding tools and for the most part, they've been pretty crappy. It's faster to just refer to StackOverflow than to deal with the hallucinations that the AI comes up with.

One tool that I have been pretty impressed with is Claude 3.5 by Anthropic. I've asked it to code a variety of tasks. First, some prediction models and while it didn't produce a top score, it produced a decent enough score on the leader board. And maybe with some better guidance, it might be able to go even higher.

Next, I got it to convert a TradingView indicator into a strategy. Unlike other models like ChatGPT and Phind, it actually managed to convert it on the first try and after optimization, got me something decent.

But the best thing so far has been a monitoring tool done in Javascript to observe a game of Crash. So, if you're familiar with the Crash game, there are probably situations where the game goes more than 5 times under 2.00x. And then the probability of hitting a 2.00x+ round increases. This way, I don't need to stare at the screen all the time and can just wait for the alert and place my bet.

Anyway, I'll be testing more scenarios and see how it goes. For the first time, AI has improved my productivity with coding.